Only a few parts of the brain appear utterly essential for human speech. Damage to these few, unique areas is therefore devastating. When patients develop focal lesions like brain tumors or epileptic foci that must be surgically removed, great care must be taken to avoid injuring nearby, normally functioning speech areas. The gold-standard technique by which speech areas are preserved during neurosurgical procedures is direct electrical stimulation of the brain. Essentially, a small probe is slowly advanced over the brain surface, delivering brief trains of electrical stimulation, while an awake patient speaks. Areas where stimulation stops speech-that is, causes speech arrest-are deemed essential for speech and left intact. Other quiet areas are thought safely removed. Yet despite how often this technique is used, we do not understand how it works. Speech arrest sites are highly variable from person to person. Their size, shape, and distribution are irregular. Yet their existence offers an important clue for the physiology underlying normal speech. Why are speech arrest sites special? Why are they located where they are? What role do they serve in normal speech production? Our research aims to answer these questions. We will use novel, high-density electrocorticography (ECoG) to record brain activity during normal speech production in patients undergoing epilepsy surgery. We have already used similar arrays to find maps of speech articulation in the human brain. We hypothesize that speech arrest sites will be found within these same articulation maps. We also hypothesize that speech arrest sites will be privileged in their functional connectivity to other speech-producing regions. Both aims will be achieved by first identifying the location of speech arrest sites in awake, behaving patients, and then determining the sites' role in articulation maps and determining their functional connectivity with other sites within the same maps. This work uses new technology to answer longstanding questions about speech. Ultimately, this work will not only help us understand the physiology of speech, but will also enable improved prediction of surgical morbidity and improved mapping of eloquent cortex prior to surgical resection. Such improvements will directly benefit patients with brain tumors, epilepsy, and disorders of language.